摘要
机械设备故障诊断中,特征融合技术能够在保留有效信息的同时去除冗余相关信息,有利于节省计算资源和提高诊断能力。提出一种基于相关性分析进行特征融合的内燃机故障诊断方法。首先,深度挖掘并提取原始信号的多域特征,形成多域特征集;然后,对多域特征集进行相关性分析,并对关联性高的特征组合进行择一保留;最后,利用主成分分析法(PCA)进行特征降维,并利用k近邻学习(kNN)算法进行故障诊断。内燃机气门间隙异常故障的有效诊断验证了该方法的适用性和准确性。
In the fault diagnosis of mechanical equipment,feature fusion technology can remove redundant relevant information while retaining valid information,which is conducive to saving computing resources and improving diagnostic capabilities.This paper proposes a fault diagnosis method for internal combustion engine based on correlation analysis and feature fusion.First,deeply mine and extract the multi-domain features of the original signal to form a multi-domain feature set;then,perform correlation analysis on the multi-domain feature set,and retain one of the feature sets with high correlation;finally,use principal component analysis method(PCA)performs feature dimensionality reduction,and uses k-nearest neighbor(KNN)learning algorithm for fault diagnosis.The effective diagnosis of the abnormal valve clearance fault of internal combustion engine verifies the applicability and accuracy of the method.
作者
张波
韩光谱
冯丞科
龚伟
周仁
ZHANG Bo;HAN Guang-pu;FENG Cheng ke;GONG Wei;ZHOU Ren(Chongqing Gas District,Petro China Southwest Oi1&Gasfield Company,Chongqing 400021,China)
出处
《机械工程与自动化》
2020年第5期128-129,132,共3页
Mechanical Engineering & Automation
关键词
内燃机
特征相关性
故障诊断
internal combustion engine
feature correlation
fault diagnosis